R Notebook

Introduction

library(ggplot2)

library(lattice)

library(latticeExtra)

library(ggpubr)

library(scatterplot3d)

library(gridExtra)

library(ggfortify)

##shiny::runGitHub("cardiomoon/ggplot2new")

One variable visualization

## Histogram in ggplot package
ggplot(data = diamonds, aes(price)) + geom_histogram()

## Histogram varied based on the cut of the diamonds (group)
ggplot(data = diamonds, aes(price, color = cut)) + geom_histogram()

## Densityplot in ggplot package
ggplot(data = diamonds, aes(price)) + geom_density()

## Densityplot varied based on the cut of the diamonds (group)
ggplot(data = diamonds, aes(price, color = cut)) + geom_density()

## Frequency plot in ggplot package
ggplot(data = diamonds, aes(price)) + geom_freqpoly()

## Frequency plot is visualized based on the cut (group)
ggplot(data = diamonds, aes(price, color = cut)) + geom_freqpoly()

## Histogram in lattice package. Price is visualized based on the cut of the diamonds. 
histogram(~ price | factor(cut), data = diamonds)

## Histogram using ggpubr package. Price is visualized based on cut of the diamonds.
gghistogram(diamonds, x = "price",
            add = "mean", rug = TRUE, color = "cut",
            palette = c("red", "blue", "green", "brown", "violet"))

Using `bins = 30` by default. Pick better value with the argument `bins`.

## Densityplots using ggpubr package. Price is visualized based on cut of the diamonds.
ggdensity(diamonds, x = "price",
          add = "mean", rug = TRUE,
          color = "cut",
          palette = c("red", "blue", "green", "brown", "violet"))

NA
NA
NA
NA

R Code for Discrete One Variable Visualization

ggplot(data = world, aes(dem_level4, fill = gdp_cap3)) + geom_bar() 

       
ggplot(data = world, aes(dem_level4, fill = regime_type3)) + geom_bar()

## Facet Grid attachment to separate based on different regions
ggplot(data = world, aes(dem_level4, fill = regime_type3)) + geom_bar() + facet_grid(regionun ~.)

R Code for Two-variable visualization

## Discrete variable (X) and continous variable (Y)
## Boxplot Visualization. Gini Coefficient (inequality) (y-variable) visualized based on regime types (x-variable). 
ggplot(data = world, aes(x = regime_type3, y = gini10, color = oecd)) + geom_boxplot() 

ggplot(data = world, aes(x = dem_level4, y =  gini10, color = oecd)) + geom_boxplot() + coord_flip()

bwplot(factor(dem_level4) ~ gini10 , data = world, xlab = "Gini Coefficient", col = "green", fill  = "blue")

## Using the ggpubr package. 
ggbarplot(data = world, x =  "dem_level4", y = "gini10", fill = "regime_type3", xlab = "Democracy Level", ylab = "Gini Coefficient", short.panel.labs= T)

## Continuous variables on both axes. Levels of gini-coefficient (y) is visualized based on democracy score (x)

ggplot(data = world, aes(x = dem_score14, y = gini10)) + geom_point()  

ggplot(data = world, aes(x = dem_score14, y = gini10, color = gdp_cap3)) + geom_point() 

ggplot(data = world , aes(x = dem_score14, y = gini10, color = gdp_cap3)) + geom_point() + facet_wrap(~regionun)

## using Lattice package 
xyplot(gini10~dem_score14 | regime_type3 , data = world, 
       type = "h", xlab = "Democracy Levels")

## Discrete variable (X) and Discrete variable (Y). Here regime type is visualized with GDP per capita. 
ggplot(data = world, aes(x = dem_level4 , y = gdp_cap3, color = regime_type3)) + geom_count()  

R Code for Correlation Relationships

##. Correlation of numerical variables
corr = cor(diamonds[, c(1, 5:7)])
ggcorrplot(corr = corr)

## using GGally package. The Bush approval ratings is used to visualize the possible correlations. 
ggpairs(approval, columns = 7:11, ggplot2::aes(color=factor(X11.Sep)))

##. R Code for Regression Relationships

##using ggplot package for regression line fit
ggplot(data = world, aes(x = dem_score14, y = gini10 )) + geom_point(na.rm = T) + geom_smooth(method = "lm", na.rm = T)  

##using ggplot package for regression line fit (loess method)
ggplot(data = world, aes(x = dem_score14, y = gini10)) + geom_point(na.rm = T) + geom_smooth(method = "loess", na.rm = T) 

##using quantile regression estimation
ggplot(data = world, aes(x = dem_score14, y = gini10)) +geom_point(na.rm = T) + geom_quantile(na.rm = T)

Regression diagnostics using ggpubr and ggfortify

## Using ggpubr package
ggscatter(data = world, x = "dem_score14", y = "gini10",
   color = "black", shape = 21, size = 3, # Points color, shape and size
   add = "reg.line",  # Add regressin line
   add.params = list(color = "blue", fill = "lightgray"), # Customize reg. line
   conf.int = TRUE, # Add confidence interval
   cor.coef = TRUE, # Add correlation coefficient. see ?stat_cor
   cor.coeff.args = list(method = "pearson", label.x = 3, label.sep = "\n")
   )

##using ggeffects, ggfortify, and performance package to estimate and diagnosize the linearmodels
##Model.1: predicting and diagonizing the generalized linear model
m1 = lm(approve~gasprice + unemploy + katrina + lrgasprice, data = approval)
autoplot(m1, which = 1:6, ncol = 3, label.size = 3)

##Model.2: predicting and diagonizing the generalized linear model with addition of iraq invasion variable
m2 = lm(approve~gasprice + unemploy + katrina + lrgasprice + iraqinvade, data = approval)
autoplot(m2, which = 1:6, ncol = 3, label.size = 3)

##Model.3: predicting and diagonizing the generalized linear model with control variables
m3 = lm(approve~gasprice + unemploy + katrina + lrgasprice + iraqinvade + lcpifood + X11.Sep, data = approval)
autoplot(m3, which = 1:6, ncol = 3, label.size = 3)

## Coefficient estimates for various models
g1 = ggcoef(m1, vline_color = "green", sort = "ascending", exclude_intercept = T)
g2 = ggcoef(m2, vline_color = "blue", sort = "ascending", exclude_intercept = T)
g3 = ggcoef(m3, vline_color = "red", sort = "ascending", exclude_intercept = T)

grid.arrange(g1, g2, g3, nrow=3)